expose model inputs and outputs with respect to model definition

[DvirDukhan]

closes #529

[codecov]

Codecov Report

Merging #552 (21b39cc) into master (bd9f462) will decrease coverage by 0.42%.
The diff coverage is 71.56%.

@@            Coverage Diff             @@
##           master     #552      +/-   ##
==========================================
- Coverage   72.57%   72.14%   -0.43%     
==========================================
  Files          35       35              
  Lines        5626     5824     +198     
==========================================
+ Hits         4083     4202     +119     
- Misses       1543     1622      +79     

Impacted Files	Coverage Δ
src/model.c	77.64% <ø> (-0.26%)	⬇️
src/backends/tflite.c	63.81% <60.86%> (-4.71%)	⬇️
src/backends/onnxruntime.c	60.75% <72.09%> (+0.57%)	⬆️
src/libtflite_c/tflite_c.cpp	58.17% <73.33%> (-0.82%)	⬇️
src/libtorch_c/torch_c.cpp	57.52% <74.35%> (+2.18%)	⬆️
src/backends/torch.c	69.86% <75.51%> (+0.37%)	⬆️
src/backends/tensorflow.c	68.82% <100.00%> (+0.17%)	⬆️
src/command_parser.c	86.66% <100.00%> (-2.06%)	⬇️
... and 3 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update bd9f462...21b39cc. Read the comment docs.

[alonre24]

Is the condition (*model)->inputs is still needed? The inputs and outputs arrays should be allocated in all backends now (not only in TF) right?

[lantiga]

Looks great, the failing tests should be just a matter of updating the tests themselves.
I would remove the condition as @alonre24 pointed out.

[lantiga]

LGTM